Networking Algorithms

1
Networking Algorithms

• Mani Srivastava UCLA Project Dynamic Sensor
  Nets (ISI-East)

2
Outline

• Dynamic location discovery
• Topology management
• Dynamic MAC address assignment

3
I. Dynamic Location Discovery

• Discovery of absolute and relative location
  important
• Location-based naming and addressing,
  geographical routing, tracking
• GPS not enough
• LOS-requirements, costly, large, power-hungry
• Ad hoc precludes trilateration with special high
  power beacons
• also, susceptible to failure
• Problem given a network of sensor nodes where a
  few nodes know their location (e.g. through GPS)
  how do we calculate the location of the other
  nodes?

4
Ad-Hoc Localization System (AHLoS)
Iterative Weighted Multilateration

• GOALS
• Localization in a distributed fashion
• Trade-offs
• Robustness
• Computation vs. communication
• Ranging using Ultrasound
• Integrated with routing messages
• Location discovery almost free
• Implementation
• Ranging using radio-synchronized ultrasound
• 3m range, noisy
• Accuracy
• Iterative 10 cm
• Collaborative 3 cm

5
Iterative Multilateration

• Atomic multilateration applied iteratively across
  the network
• may stall if network is sparse, of beacons is
  low, terrain obstacles

Resolved Nodes
Total Nodes
Initial Beacons
Uniformly distributed deployment in a field
100x100. Node range 10.
6
Iterative Multilateration Accuracy
50 Nodes 10 beacons 20mm white gaussian ranging
error
7
Collaborative Multilateration

• Step 1 form Collaborative Subtrees within
  their neighborhood
• an unknown node is collaborative if it has at
  least 3 participating neighbors
• a node is participating if it is either a beacon,
  or if it is an unknown node that is also
  participating
• at each node at least one of its participating
  nodes are new to the set
• at least one of the beacons used to determine
  the position of a node should not be collinear
  with the other beacons used to determine the node
  position
• Step II obtain initial location estimate for
  subsequent computation
• use beacon locations hop distances to obtain
  approximate location bounds
• Step III perform computation
• Measurement Update part of Kalman Filter
• Centralized, at a leader elected in the subtree
• or, Fully Distributed
• can start computing locations based on node
  connectivity and initial estimate

Uncertainty of estimated location in first
iteration
Uncertainty of estimated location in second
iteration
8
Example

• Network of 30 nodes
• 6 beacons, 24 unknowns
• Ranging noise experimentally derived

9
Computation Communication Expense
Computation Expense Results using the FLOPS
command in MATLAB
Computation vs. Communication Tradeoff

• Total number of transmissions
• Centralized 70 packets
• Distributed 416 packets
• Centralized approach has additional overhead for
  leader election

10
II. Topology Management

• Two phases of sensor network operation
• Detect event
• Relay information to users
• Energy consumption of radio dominates that of
  sensors CPU
• ? perform event detection continuously
• The only energy efficient mode of the radio is
  the sleep mode
• ? put radio to sleep as often as possible
• Existing approaches density-energy trade-off
• keep enough nodes awake to handle the data
  forwarding (forwarding state)
• but for substantial energy savings we need large
  densities
• Observation
• most of the time, the network is only monitoring
  its environment, waiting for an event to happen
  (monitoring state)
• Idea
• put node radios to sleep and wake them up when
  they need to forward data
• low duty-cycle paging channel using a 2nd radio
  trades off energy savings for setup latency

11
STEM High-level Operation
Wakeup plane
Power
f1
Tx
Time
Power
Data plane
f2
Tx /Rx
Sleep
Initiator node
Target node
Rx
Wakeup plane
Power
f1
Sleep
Time
Power
Data plane
f2
Tx /Rx
Sleep
12
Detailed Operation
Initiator node
f1
B1
B2
1. beacon received
Train of beacon packets
TRx
2. beacon acknowledge
T
f1
Target node
13
Latency Energy Analysis
Wakeup plane
f1
Data plane
f2
Forwarding state
Monitoring state
Fraction of time in the forwarding state ?

• Setup latency
• Energy savings

Appropriate choice of interval sizes
Mostly monitoring state ? ltlt 1 or ? gtgt 1
14
Energy-Latency Trade-off
? 101
? 102
TRx 0.225 s
? 103
? 104

• The tradeoff between energy and delay is
  manipulated by varying T
• T ? ? E ? TS ?
• The energy savings increase as the monitoring
  state becomes more dominant, ? ?

15
Topology Management in Forwarding State
GAF Geographic Adaptive Fidelity Ya2001

• Conserve traffic forwarding capacity
• Divide network in virtual grids
• Each node in a grid is equivalent from a traffic
  forwarding perspective
• Keep 1 node awake in each grid at each time

• GAF reduces the energy by a factor M
• This factor is a function of the average number
  of nodes in a grid M

Average number of neighbors of a node
for uniformly random node deployment
16
Comparing STEM GAF
STEM
Curve of comparable energy savings
Leverage latency
?
Leverage density
GAF
17
Combining STEM and GAF forJoint
Energy-Latency-Density Trade-off

• As in GAF, 1 node is active in each grid
• ? the grid can be considered a virtual node
• This virtual node runs the STEM protocol

STEM alone
? 10
GAF alone
? 30
? 60
? 100
? 200
18
III. Dynamic MAC Address Assignment

• Wireless spectrum is broadcast medium
• MAC addresses are required
• In wireless sensor networks, data size is small
• Unique MAC address present unneeded overhead
• Employ spatial address reuse (similar to reuse in
  cellular systems)
• MAC address, link ids
• Two aspects
• Dynamic assignment algorithm
• Address representation

19
Distributed Assignment Algorithm

• Network is operational (nodes have valid
  address)
• Listen to periodic broadcasts of neighboring
  nodes
• In case of conflict, notify node
• (this node resends a broadcast)
• Choose non-conflicting address and broadcast
  address in a periodic cycle. At this point the
  new node has joined the network.
• Additive convergence network remains operational
  during address selection
• Mapping unique ID to spatially reusable address
• Algorithm also valid when unidirectional links

20
Encoded Address Representation

• Size of the address field?
• Non-uniform address frequency
• Huffman encoding
• Robust can represent any address
• Practical address selection
• All addresses with same codeword size are
  equivalent
• Choose random address in that range to reduce
  conflict messages

21
Non-uniform Network Density
22
Effect of Packet Losses (? 10)
23
Scalability

• Address assignment
• Distributed algorithm with periodic localized
  communication
• Address representation
• Encoded addresses depend only on distribution

Scales perfectly (neglecting edge effects)
Assignment
Representation
24
Simulation Results
Fixed size dynamic
Our schemes
25
Dynamic Address Allocation Summary

• Spatial reuse of address
• Dynamic assignment algorithm
• Localized scalability
• Additive convergence robustness
• Encoded address representation
• Independent of network size scalability
• Variable length addresses robustness